comp.lang.ada
 help / color / mirror / Atom feed
* Re: GNAT: Storage size?
  1996-07-04  0:00 GNAT: Storage size? Joerg Rodemann
@ 1996-07-04  0:00 ` Robert Dewar
  1996-07-05  0:00   ` Jim Easterbrook
  1996-07-06  0:00 ` Laurent Guerby
  1 sibling, 1 reply; 4+ messages in thread
From: Robert Dewar @ 1996-07-04  0:00 UTC (permalink / raw)



Joerg said

"I am doing some matrix diagonalization using Ada-95 for the big frame of
my programs and Fortran-77 for most of the matrix routines. Currently I am
working under HP UX 9.05 (we're upgrading to 10.10 during the next months,
I hope...). Since my matrices become quite large (at the moment 900x900
fortran double precision) I encountered a problem with GNAT. I think there
is a simple solution to the problem, but have no idea were to look for it:"

Without the slightest idea of what you are doing in detail, neither do we
(have any idea where to look for it). It's possible you are exceeding
stack limits, but without sample source code it is impossible to guess
what might be wrong. Try to narrow down a simple example of something
that does not work, and send it along to report@gnat.com. This might
be a  bug in GNAT, but I doubt it, it is unlikely that there is a bug
selectively affecting large arrays, but anyway, without source, we cannot
do anything more.





^ permalink raw reply	[flat|nested] 4+ messages in thread

* GNAT: Storage size?
@ 1996-07-04  0:00 Joerg Rodemann
  1996-07-04  0:00 ` Robert Dewar
  1996-07-06  0:00 ` Laurent Guerby
  0 siblings, 2 replies; 4+ messages in thread
From: Joerg Rodemann @ 1996-07-04  0:00 UTC (permalink / raw)



Hello folks!

I am doing some matrix diagonalization using Ada-95 for the big frame of
my programs and Fortran-77 for most of the matrix routines. Currently I am
working under HP UX 9.05 (we're upgrading to 10.10 during the next months, 
I hope...). Since my matrices become quite large (at the moment 900x900
fortran double precision) I encountered a problem with GNAT. I think there
is a simple solution to the problem, but have no idea were to look for it:

   Trying to run the program with still larger matrices my program ends up
   with a Segmentation Fault quite at the beginning. My guess is that the
   Ada RTS only supplies memory for my data up to an upper limit. Any
   suggestions how this can be overcome? (My aim is a matrix of 1800x1800
   fortran double precision.) I am not using any tasks, just some generics
   and the Interface.Fortran libraries. Sure I also use the usual stuff as
   Ada.Text_Io and so on.

Any help is appreciated

Thanks

George Rodemann

P.S.: A copy of this posting is sent to report@gnat.com

--
Joerg 'George' Rodemann                      Erhard-Groezinger-Strasse 82
Department of Physics                                   D-89134 Blaustein
University of Ulm                                      Tel. (0731) 553319
e-mail:                                    rodemann@mathematik.uni-ulm.de




^ permalink raw reply	[flat|nested] 4+ messages in thread

* Re: GNAT: Storage size?
  1996-07-04  0:00 ` Robert Dewar
@ 1996-07-05  0:00   ` Jim Easterbrook
  0 siblings, 0 replies; 4+ messages in thread
From: Jim Easterbrook @ 1996-07-05  0:00 UTC (permalink / raw)



In article <dewar.836481757@schonberg>, Robert Dewar (dewar@cs.nyu.edu) wrote:
: Joerg said
: 
: "I am doing some matrix diagonalization using Ada-95 for the big frame of
: my programs and Fortran-77 for most of the matrix routines. Currently I am
: working under HP UX 9.05 (we're upgrading to 10.10 during the next months,
: I hope...). Since my matrices become quite large (at the moment 900x900
: fortran double precision) I encountered a problem with GNAT. I think there
: is a simple solution to the problem, but have no idea were to look for it:"
: 
: Without the slightest idea of what you are doing in detail, neither do we
: (have any idea where to look for it). It's possible you are exceeding
: stack limits, but without sample source code it is impossible to guess
: what might be wrong. Try to narrow down a simple example of something
: that does not work, and send it along to report@gnat.com. This might
: be a  bug in GNAT, but I doubt it, it is unlikely that there is a bug
: selectively affecting large arrays, but anyway, without source, we cannot
: do anything more.

This sounds very like a problem I encountered when using GNAT for
picture processing programs on SunOS. I was getting segmentation
violation errors when processing large pictures. The cure was to set a
larger stack limit with "ulimit".
-- 
Jim Easterbrook                          <jim.easterbrook@rd.bbc.co.uk>
  Senior R & D Engineer, Studio Section     |         PGP key available
  BBC Research & Development                |    Phone: +44 1737 836530
  Kingswood Warren, Tadworth                |    Fax:   +44 1737 836665
  Surrey  KT20 6NP, UK                      |
*** All opinions are mine and are not necessarily shared by the BBC ***




^ permalink raw reply	[flat|nested] 4+ messages in thread

* Re: GNAT: Storage size?
  1996-07-04  0:00 GNAT: Storage size? Joerg Rodemann
  1996-07-04  0:00 ` Robert Dewar
@ 1996-07-06  0:00 ` Laurent Guerby
  1 sibling, 0 replies; 4+ messages in thread
From: Laurent Guerby @ 1996-07-06  0:00 UTC (permalink / raw)



Joerg> [... Fortran ...]

   There's a binding to LAPACK in
ftp://cs.nyu.edu/pub/gnat/contrib. this won't solve your stack
overflow problem, but it may be of some help ;-). You should also
consider declaring your big arrays at library level.

-- 
Laurent Guerby <guerby@gnat.com>, Team Ada.
   "Use the Source, Luke. The Source will be with you, always (GPL)."




^ permalink raw reply	[flat|nested] 4+ messages in thread

end of thread, other threads:[~1996-07-06  0:00 UTC | newest]

Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
1996-07-04  0:00 GNAT: Storage size? Joerg Rodemann
1996-07-04  0:00 ` Robert Dewar
1996-07-05  0:00   ` Jim Easterbrook
1996-07-06  0:00 ` Laurent Guerby

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox